AI Paper Atlas

Research map / Label-efficient and robust learning / Semi-supervised learning

Semi-supervised Learning with a Teacher-student Network for Generalized Attribute Prediction

Minchul Shin

ECCV 2020 · Label-efficient and robust learning › Semi-supervised learning · approach: Confidence-based pseudo-label selection · 14 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add focal term (gamma=1) to teacher loss (ablation)Ours(T) · DeepFashion CategoryR@3: 88.45 -> 89.32Improved
tune gamma per attribute type for teacher (ablation)Ours(T) · DeepFashion CategoryR@3: 89.32 -> 90.01Improved
use label embedding instead of binary cross-entropy (ablation)Ours teacher · FiccYF1@1 avg: +0.62%Improved
distill student from teachersOurs(S) vs Ours(T) · iMatFashion GenderF1@1: 88.50 -> 90.46Improved
distill student from teachersOurs(S) vs Ours(T) · iMatFashion CategoryF1@1: 42.50 -> 42.43No clear effect
distill student from teachersOurs(S) vs Ours(T) · FiccY TextureF1@1: 57.74 -> 59.19Improved
supervise student with four teachers instead of one (ablation)Ours(S) · FiccY TextureR@3: 93.57 -> 94.22Improved
supervise student with four teachers instead of one (ablation)Ours(S) · FiccY CategoryR@3: 97.30 -> 97.30No clear effect
add unlabeled target-domain images (DARN) to student trainingOurs(S) · DARN cross-domain, DeepFashion trainF1@1 All: 65.15 -> 65.49Improved
add unlabeled target-domain images (DARN) to student trainingOurs(S) · DARN cross-domain, iMatFashion trainF1@1 All: 73.10 -> 73.22Improved
mix 1M corrupted images into student trainingOurs(S) · DeepFashion corruptedR@3: 42.96 -> 43.96Improved
mix corrupted images into student trainingOurs(S) · DeepFashion cleanR@3: 57.47 -> 57.80Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy